About a month ago I submitted the following problem:
We have a group of sparcstations running SUNOS 4.1.3 and Oracle 7.1.2 .
There are two oracle servers with separate databases. The sparcstation
users may access databases on either server from time to time. The oracle
parameters in the initoracle.ora file are similar on both servers.
The servers are configured:
Server A:
Sparc10, SUNOS 4.1.3, Oracle 7.1.2, Sqlnet V1
Server B:
SS1000E, Solaris 2.4, Oracle 7.1.4, Sqlnet V1
The Problem:
Often, when a COMMIT is executed for a 'large' number of data items
on Server B, the client sparcstation's oracle process will hang.
There is no error msg. to the client and no error recorded by
the server in the alert_oracle.log file or in the .trc trace files.
The identical COMMIT will never fail when executed on Server A.
I can't really quantify 'large' number of items. Users say that
about 200 items can always commit successfully but failures begin
when 300 or 400 items are committed.
SOLUTION:
I received a couple of suggestions, no solutions, and nobody else seems to have
observed this problem. Oracle's support folks offered no help. A few days ago, in
the Sunsolve database of bug reports, I found BugID 1187901 which describes Oracle
processes waiting forever to acquire an internal oracle latch. This BugID is
fixed in jumbo kernel patch 101945-27. We installed the patch and it fixed the
problem. So, if you're running Oracle 7 under Solaris 2.4, it is a very good idea
to install jumbo kernel patch 101945-27 or later.
Thanks to the following, and any who may have been omitted:
abbas@houston.Geco-Prakla.slb.com (Abbas Mehrabian )
jckelly@netcom.com (John Kelly)
Bob Reardon
bob.reardon@SLB.COM
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:10:32 CDT